Exercise: Assembly Validation

Questions

If you have questions about the lab after the course, you are welcome to contact me: martin.norling@nbis.se

Gene Space

Finding genes and other genomic regions is generally the domain of annotation, not assembly – we can cheat a bit though! Making a rough estimate of the gene content compared to what is expected from a complete assembly can be a hint at the state of the assembly (at least of the assembled gene space).

BUSCO

Busco, Benchmarking Universal Single-Copy Orthologs, is a program that attempts to estimate how well an assembly covers the gene space of a particular linage. Load the BUSCO module with:

module load bioinfo-tools BUSCO/1.22

There will be some information printed when you load the BUSCO module - this information is really important so it’s a good idea to read it! Run BUSCO –h to get information on how to run. You will want to run Busco using -l $BUSCO_LINEAGE_SETS/bacteria to look for bacteria, and -c 8 to use 8 threads.

When the program finishes, look through the output directory for the files full_table_*, missing_buscos_list_*, and short_summary_*.

• Does the assembly seem to include the full gene set based on these results?

Whole Genome Alignment

When a reference is available, we can use whole genome alignment to compare the assemblies and look for differences and similarities.

MUMmer

The program we will use to look do whole genome alignment is called nucmer and is from the MUMmer package. To use nucmer, load the MUMmer/3.23 module. Running nucmer is fairly straight forward, type nucmer -h to get some help, and then run with your assembly, the provided reference from yesterday (from the illumina assembly part) and use the -p flag to set output name.

This will produce a mapping between the reference and the assembly, but it won’t be sorted, so the output will be very noisy. You can use the show-tiling command to sort the delta file like this:

show-tiling [filename].delta > [filename].tiling

Dotplots

There is a built-in tool called mummerplot that can help us plot the output. It will default to trying to render directly to X11 though, so run with the --png flag to write an image file. As with nucmer you should also use the -p flag to set output name. Run mummerplot on the delta-file that nucmer produced, as well as the tiling file. Download both files and have a look!

• Does your assembly match well to the reference?
• Did the plots look like you thought they would?

The mummerplot of a tiling dataset isn’t the same kind of plot as the original dotplot. To remedy this, we’ve also supplied a simple python script to make dotplots with. The script is called dotplot.py and it can read .coord and .tiling files. It is available in /proj/g2016024/nobackup/illumina_assembly/scripts/dotplot.py, and you need to load the biopython module for it to run.

Mauve

For the final part we will use a program called Mauve, that you’ll have to run on your own machine! Start by downloading Mauve for your operating system from the Mauve website.

You will also need to download the assemblies that you wish to look at, as well as the reference to you local machine using scp.

Start Mauve, and select File -> Align with ProgressiveMauve, and select the sequences that you wish to align. You can give it multiple sequences, but I’d recommend doing pairwise alignments. Once the alignment is done you can run Tools -> Move Contigs to have mauve recursively move the contigs around to match better each other better.